# To help us fund GROMACS development, we humbly ask that you cite
# the research papers on the package. Check out http://www.gromacs.org.
+include(CheckCCompilerFlag)
+include(CheckCXXCompilerFlag)
+
# Test C flags FLAGS, and set VARIABLE to true if the work. Also add the
# flags to CFLAGSVAR.
MACRO(GMX_TEST_CFLAG VARIABLE FLAGS CFLAGSVAR)
# Prepare some local variables so CUDA and non-CUDA code in targets
# works the same way.
function(gmx_target_compile_options_inner)
- set (CFLAGS "${SIMD_C_FLAGS};${MPI_COMPILE_FLAGS};${EXTRA_C_FLAGS};${GMXC_CFLAGS}" PARENT_SCOPE)
+ set(CFLAGS
+ ${SIMD_C_FLAGS}
+ ${MPI_C_COMPILE_OPTIONS}
+ ${EXTRA_C_FLAGS}
+ ${GMXC_CFLAGS}
+ PARENT_SCOPE)
+
# When SYCL support has been enabled (so the flag is non-empty), we still *disable* things
# by default to avoid running each file three passes through the compiler. Then we'll explicitly
# enable SYCL for the few files using it, as well as the linker.
- set (CXXFLAGS "${SIMD_CXX_FLAGS};${MPI_COMPILE_FLAGS};${DISABLE_SYCL_CXX_FLAGS};${EXTRA_CXX_FLAGS};${GMXC_CXXFLAGS}" PARENT_SCOPE)
+ set(CXXFLAGS
+ ${SIMD_CXX_FLAGS}
+ ${MPI_CXX_COMPILE_OPTIONS}
+ ${DISABLE_SYCL_CXX_FLAGS}
+ ${EXTRA_CXX_FLAGS}
+ ${GMXC_CXXFLAGS}
+ PARENT_SCOPE)
endfunction()
# Implementation function to add compiler flags expected for all
if(NOT GMX_OPENMP)
GMX_TEST_CXXFLAG(CXXFLAGS_PRAGMA "-Wno-unknown-pragmas" GMXC_CXXFLAGS)
endif()
+ GMX_TEST_CXXFLAG(CXXFLAGS_WARN_NO_RESERVED_IDENTIFIER "-Wno-reserved-identifier" GMXC_CXXFLAGS) # LLVM BUG #50644
GMX_TEST_CXXFLAG(CXXFLAGS_WARN_NO_MISSING_FIELD_INITIALIZERS "-Wno-missing-field-initializers" GMXC_CXXFLAGS)
- # Intel LLVM 2021.2 defaults to no-finite-math which isn't OK for GROMACS
- if(GMX_INTEL_LLVM AND GMX_INTEL_LLVM_VERSION GREATER_EQUAL 2021020)
- GMX_TEST_CXXFLAG(CXXFLAGS_FINITE_MATH "-fno-finite-math-only" GMXC_CXXFLAGS)
- endif()
- # Some versions of Intel ICPX compiler (at least 2021.1.1 to 2021.2.0) fail to unroll a loop
+ # Some versions of Intel ICPX compiler (at least 2021.1.1 to 2021.3.0) fail to unroll a loop
# in sycl::accessor::__init, and emit -Wpass-failed=transform-warning. This is a useful
# warning, but mostly noise right now. Probably related to using shared memory accessors.
# Note: not a typo: ICPX 2021.1.1 has GMX_INTEL_LLVM_VERSION 202110; 2021.2.0 has 20210200.
endif()
endmacro()
+
+# Make sure we generate warnings (and hopefully fix) "everything"
+# reported by compilers that support that (ie recent clang and its
+# derivatives).
+function(gmx_warn_on_everything target)
+
+ # If the compiler suports warning on "everything" then we'll turn
+ # it on. Note that all warnings become errors for developer
+ # builds, but not for user builds.
+ gmx_target_warning_suppression(${target} "-Weverything" HAS_WARNING_EVERYTHING)
+
+ if (NOT HAS_WARNING_EVERYTHING)
+ # There's no need to suppress aspects of "-Weverything" if
+ # that warning is not supported.
+ return()
+ endif()
+
+ # We don't actually fix everything, so list the exceptions that we
+ # choose to make. We may be able to eliminate some of these over
+ # time.
+ #
+ # We check whether the flag is accepted first, so that we suppress
+ # such warnings also with compilers that don't directly identify
+ # as e.g. clang despite being based on it (e.g. most vendor
+ # compilers), and also don't fail to compile GROMACS when future
+ # versions of any such compiler changes how the warnings
+ # look/work.
+
+ # We have no intention of C++98 compability
+ gmx_target_warning_suppression(${target} "-Wno-c++98-compat" HAS_WARNING_NO_CPLUSPLUS98_COMPAT)
+ gmx_target_warning_suppression(${target} "-Wno-c++98-compat-pedantic" HAS_WARNING_NO_CPLUSPLUS98_COMPAT_PEDANTIC)
+
+ # Don't warn for use of OpenMP pragmas in no-omp build
+ gmx_target_warning_suppression(${target} "-Wno-source-uses-openmp" HAS_WARNING_NO_SOURCE_USED_OPENMP)
+
+ # Allowed in attributes (compilers are required to ignore unknown attributes)
+ gmx_target_warning_suppression(${target} "-Wno-c++17-extensions" HAS_WARNING_NO_CPLUSPLUS17_EXTENSIONS)
+
+ # Custom Doxygen commands are used
+ gmx_target_warning_suppression(${target} "-Wno-documentation-unknown-command" HAS_WARNING_NO_DOCUMENTATION_UNKNOWN_COMMAND)
+
+ # We need to use default labels in switch statements, because GCC gives
+ # maybe-uninitialized without default label and checks for illegal enum values.
+ gmx_target_warning_suppression(${target} "-Wno-covered-switch-default" HAS_WARNING_NO_COVERED_SWITCH_DEFAULT)
+
+ # Default statement for enum is OK.
+ # It's OK to not have branches for Count members of enum classes
+ gmx_target_warning_suppression(${target} "-Wno-switch-enum" HAS_WARNING_NO_SWITCH_ENUM)
+
+ # We need to use macros like
+ # GMX_CATCH_ALL_AND_EXIT_WITH_FATAL_ERROR and
+ # CLANG_DIAGNOSTIC_IGNORE. Those will look strange if they don't
+ # have a semicolon after them, and might confuse tools like IDEs
+ # also.
+ gmx_target_warning_suppression(${target} "-Wno-extra-semi-stmt" HAS_WARNING_NO_EXTRA_SEMI_STMT)
+
+ # We intend to use fully inline classes with virtual methods
+ gmx_target_warning_suppression(${target} "-Wno-weak-vtables" HAS_WARNING_NO_WEAK_VTABLES)
+
+ # We intend to use constructor arguments that shadow member variables
+ gmx_target_warning_suppression(${target} "-Wno-shadow" HAS_WARNING_NO_SHADOW)
+
+ # Padding of structs is routine, we don't need to hear about it
+ gmx_target_warning_suppression(${target} "-Wno-padded" HAS_WARNING_NO_PADDED)
+
+ # Our uses of double underscores in macro names are OK
+ gmx_target_warning_suppression(${target} "-Wno-reserved-id-macro" HAS_WARNING_NO_RESERVED_ID_MACRO)
+
+ # Implicit conversion of float to double is fine
+ gmx_target_warning_suppression(${target} "-Wno-double-promotion" HAS_WARNING_NO_DOUBLE_PROMOTION)
+
+ # No resources in static variables need exit-time destructors
+ gmx_target_warning_suppression(${target} "-Wno-exit-time-destructors" HAS_WARNING_NO_EXIT_TIME_DESTRUCTORS)
+
+ # Global constructors are not needed
+ gmx_target_warning_suppression(${target} "-Wno-global-constructors" HAS_WARNING_NO_GLOBAL_CONSTRUCTORS)
+
+ # False positives are emitted
+ gmx_target_warning_suppression(${target} "-Wno-documentation" HAS_WARNING_NO_DOCUMENTATION)
+
+ # We intend to use format strings that we construct, even though that is a security risk
+ gmx_target_warning_suppression(${target} "-Wno-format-nonliteral" HAS_WARNING_NO_FORMAT_NONLITERAL)
+
+ # We do a lot of conditional compilation that sometimes uses a symbol and sometimes does not
+ gmx_target_warning_suppression(${target} "-Wno-used-but-marked-unused" HAS_WARNING_NO_USED_BUT_MARKED_UNUSED)
+
+ # It's only risky to compare floats for equality when they are the
+ # result of computation. Unfortunately it's hard to tell the
+ # difference and there's no good way to suppress this on a
+ # case-by-base basis.
+ gmx_target_warning_suppression(${target} "-Wno-float-equal" HAS_WARNING_NO_FLOAT_EQUAL)
+
+ #
+ # Exceptions we should consider fixing
+ #
+
+ # Much code in gmxana uses complex logic that may or may not be valid
+ gmx_target_warning_suppression(${target} "-Wno-conditional-uninitialized" HAS_WARNING_CONDITIONAL_UNINITIALIZED)
+
+ # We have many places implicit conversions still occur, most of which need fixing
+ gmx_target_warning_suppression(${target} "-Wno-conversion" HAS_WARNING_NO_CONVERSION)
+
+ # We use the Linux signal handlers in the intended way, but it triggers this warning.
+ # It would be better to localize this exception.
+ gmx_target_warning_suppression(${target} "-Wno-disabled-macro-expansion" HAS_WARNING_NO_DISABLED_MACRO_EXPANSION)
+
+ # The NBNXM simd kernels define lots of macros that are not used
+ # It would be better to localize this exception.
+ gmx_target_warning_suppression(${target} "-Wno-unused-macros" HAS_WARNING_NO_UNUSED_MACROS)
+
+endfunction()
biod.pnpi.spb.ru / alexxy / gromacs.git / blobdiff